Using hazard information for establishing a rationale decision ...The principles of a rationale...

Using hazard information for establishing a

rationale decision-making

Warner Marzocchi Istituto Nazionale di Geofisica e Vulcanologia

Understanding and managing induced seismicity Davos, 11-13 March, 2015

Outline !  Introducing the problem: the separation of roles in the

decision making process

!  Hazard/risk separation principle

!  The principles of a rationale decision making; when it is important to make a distinction of roles.

!  Some real cases, where this distinction is not applied

!  Risk reduction process requires different expertise (scientific and non-scientific), and the boundaries between these expertise are often fuzzy. Often scientists act unconsciously as decision makers.

!  Separating and clarifying roles is important! It is not only matter of culpability. This clear separation "  allows each partner to protect the integrity of their specific

assessment; "  clarifies the competences required at each step of the risk

reduction process; "  facilitates the establishment of transparent and clear decision

making protocols

!  Hazard/risk separation principle is important to make this separation

Introducing the problem

Introducing the problem

The Hazard/Risk separation principle

Expert judgment

Physics-based models

Empirical models

Probabilistic forecasting information

data & observations

Hazard/Risk

interface (definition of a proper format of

the hazard for each

end users)

User 1 Risk Analysis (definition of a

risk metric)

User 2 Risk Analysis (definition of a

risk metric)

User N Risk Analysis (definition of a

risk metric)

User 1 Mitigation

Options (defining the set of options

and evaluating the effect)

User 2 Mitigation



User N Mitigation



User 1 Decision

making (selecting the most useful

mitigation options)

User 2 Decision


mitigation options)

User N Decision


mitigation options)

Hazard&forecas,ng&domain& Risk&management&domain&

…………..&

Hazard analysis is purely driven by Science. Risk analysis and mitigation is more heterogeneous and Science is not enough (different levels of decision-making)

Hazard/Risk separation principle

Notwithstanding any scientist knows very well the distinction of hazard and risk, we note that the hazard/risk separation principle is not often properly acknowledged by many scientists working in hazard analysis (that sometimes tend to be overconfident on their capability to reduce the risks for society) Some examples… !  Scientists that define which event probability (hazard) is negligible or

not (Operational Earthquake Forecasting)

!  Scientists that advocate the (worst) scenario to be used in risk mitigation

!  Scientists that define alert levels in volcanic systems

!  Scientists that define a traffic light for induced seismicity

!  Hazard/Risk separation principle

The Principles of rationale decision making

The principles of a rationale decision-making

Evacuate

Weigh the pros and cons of evacuation

Evacuate

Evacuate

Traditional deterministic precautionary approach for Civil Protection

If there is a potential public danger, a precautionary evacuation would prioritize safety above any other considerations.

No risk assessment is needed for most evacuation decisions

Hurricane

Remote large tsunami

Tornado

Lava flow

LOW HIGH

HIGH

LOW

Bomb alert

(Small) flank collapse at Stromboli

Explosive eruption in high risk volcanoes

Likelihood of false alarm

Num

ber o

f eva

cuee

s


Quantitative Risk assessment is particularly important in some situations

Hurricane

Remote large tsunami

Tornado

Lava flow

LOW HIGH

HIGH

LOW

Bomb alert

(Small) flank collapse at Stromboli

Explosive eruption in high risk volcanoes

Likelihood of false alarm

Num

ber o

f eva

cuee

s


Quantitative Risk assessment is particularly important in some situations

Weigh the pros and cons


Recently one decision maker told me: “If you want to make a separation of roles you have to give me probabilities: otherwise, please let me know also what I have to do… but forget any distinction in roles” (and responsibilities)

Evacuate

Evacuate


Evacuate

Weigh the pros and cons of evacuation

Evacuate

Evacuate

The Challenge is for scientists to articulate uncertainty without losing credibility and to give public officials the information they need for decision-making

Scientists Public officials

this requires to bridge the gap between scientific output (probability) and the boolean logic (YES-NO) of decision-makers

ACTION

Adverse Hazard State

Not Adverse Hazard State [a] Take action

C

C [b] do NOT take action

L

0

Loss-Cost Matrix


Cost-benefit analysis of precautionary mitigation action (1)

C is the cost if a mitigation action is taken. P * L is the cost if a mitigation action is not taken.

If P * L > C, the cost for society “probably” lost exceeds the cost of the mitigation action. Therefore, the mitigation action should be taken when

P > C / L



P > C / L BASIC ROLE FOR MAKING A DECISION UNDER UNCERTAINTY

Scientific domain(hazard)

Risk analysis and decision-makers domain

C is the cost if a mitigation action is taken. P * L is the cost if a mitigation action is not taken.

If P * L > C, the cost for society “probably” lost exceeds the cost of the mitigation action. Therefore, the mitigation action should be taken when



Some cases where hazard/risk separation principle is not acknowledged

Evolution of the weekly probability with time for the selected area: updated every day or after a M3.5+

Saying that the probability of an earthquake is negligible

!  Seismic (and risk) hazard varies with time (in particular in the short-term)

!  During a seismic sequence the weekly probability of a destructive earthquake can increase 100-1000 times with respect to the reference level (derived from the long-term hazard), but this probability rarely reaches 1%. (NOTE: OEF does not necessarily imply ‘small’ probabilities)

!  Some models based on earthquake clustering provide accurate estimations of such probabilities (continuously under test through CSEP experiments)

!  Despite the usual belief, such models are verified empirically much better than long-term hazard models.

The Science of OEF



… you can’t issue any alert/warning with such probabilities; such probabilities (up to 5%) are not useful for saving lives [Wang & Rogers, 2014]


Weekly Individual risk of death less than 10 km from seismic sequence (in parenthesis annual risk)

Jan, 1, 2010 Oct, 25, 2012 Oct, 26, 2012 Jul, 21, 2013

No seismic sequence ongoing

Seismic sequence ongoing


2 10^-7 (10^-5)

Jan, 1, 2010 Oct, 25, 2012 Oct, 26, 2012 Jul, 21, 2013





2 10^-8 (10^-6)

2 10^-7 (10^-5)

2 10^-6 (10^-4)

Jan, 1, 2010 Oct, 25, 2012 Oct, 26, 2012 Jul, 21, 2013





Weekly Prob. M5.5+ (in parenthesis)

2 10^-8 (10^-6)

2 10^-7 (10^-5)

2 10^-6 (10^-4)

(0.0018%)

Jan, 1, 2010 Oct, 25, 2012 Oct, 26, 2012 Jul, 21, 2013





Weekly Prob. M5.5+ (in parenthesis)

2 10^-8 (10^-6)

2 10^-7 (10^-5)

2 10^-6 (10^-4)

(0.7%)

(0.0018%)

Jan, 1, 2010 Oct, 25, 2012 Oct, 26, 2012 Jul, 21, 2013




Some scientists advocate the need to protect society from the worst scenario. This sounds very appealing (assuming it is possible to define the ‘worst’), but it does not lead to ethical and rationale decision-making.

Choosing the (worst) scenario

Some scientists advocate the need to protect society from the worst scenario. This sounds very appealing (assuming it is possible to define the ‘worst’), but it does not lead to ethical and rationale decision-making. However, money for risk reduction are bounded and we cannot reduce all risks to zero. Need a holistic view of all risks


Some scientists advocate the need to protect society from the worst scenario. This sounds very appealing (assuming it is possible to define the ‘worst’), but it does not lead to ethical and rationale decision-making. However, money for risk reduction are bounded and we cannot reduce all risks to zero. Need a holistic view of all risks Example: A M6.5 below London is possible. Should the UK government retrofit the whole city? Or, is it better to spend these money to protect the city from the floods of Thames that will likely increase due to the climate change?


Some scientists advocate the need to protect society from the worst scenario. This sounds very appealing (assuming it is possible to define the ‘worst’), but it does not lead to ethical and rationale decision-making. However, money for risk reduction are bounded and we cannot reduce all risks to zero. Need a holistic view of all risks Example: A M6.5 below London is possible. Should the UK government retrofit the whole city? Or, is it better to spend these money to protect the city from the floods of Thames that will likely increase due to the climate change? Example: Should we protect Naples from the worst scenario of Campi Flegrei? The products of the last Campi Flegrei eruption arrived close to Rome. Should we plan an evacuation of more than 10 millions of people? Or should we take into account that the most likely eruption is of much smaller magnitude?


Defining the alert systems

Each color corresponds to a set of specific actions that have to be made by Civil Protection. Moving from one color to another means a lot of things in terms of impact on society. So, how can volcanologists define ‘low’, ‘medium’ or ‘high’ probability? These thresholds do not have any scientific meaning, but they have a lot to do with the costs and benefits of any set of mitigation actions.

Defining the traffic light

Usually, the traffic lights are defined according to the magnitude of the event. So, it is implicitly assumed that the higher the magnitude, the higher the risk for the future; but there is no any quantitative estimation. In practice, the threshold magnitudes are not defined according to pure scientific thoughts, but looking at the costs to keep the traffic light amber (or red) too often.

Few final remarks

!  The hazard/separation principle is essential to separate and clarify roles and responsibilities in the risk reduction process.

!  Such a distinction facilitates the interaction of different experts in planning transparent risk reduction protocols.

!  This principle is very important for governmental institution to define their mission.

!  A single scientist can wear different hats simultaneously. S/he just need to be aware that each hat requires (very) different competences, and that a good, even an excellent scientist, is not necessarily a good decision-maker

Thank you

Using hazard information for establishing a rationale decision ...The principles of a rationale...

Documents

Transcript of Using hazard information for establishing a rationale decision ...The principles of a rationale...